Organized By
CEE Seminar Committee
Host By
Associate Professor YI Yaolin
Topic
Data-efficient Modelling of Soil Properties with Uncertainty Quantification
About the Seminar
This study aims to achieve efficient modelling of soil properties through effective uncertainty quantification, data fusion, and automated data acquisition. First, we apply uncertainty quantification techniques to various deterministic machine learning algorithms to enhance their reliability and interpretability. Second, we employ a multi-precision learning approach, fusing high-precision and low-precision data from different sources to reduce the amount of data required to
predict soil properties in a specific region. Building on this, we further combine an uncertainty-based active learning strategy with multi-precision learning to guide data acquisition and further reduce data requirements.
This approach offers significant advantages compared to methods without data fusion and automated data acquisition. Finally, we develop a general-purpose machine learning modelling platform with a user-friendly graphical interface. This platform allows users to complete all operations—data import, pre-processing, algorithm selection, hyperparameter optimization, model building, model evaluation, model storage, new data application, and result visualization—with a simple click, thus simplifying the machine learning modelling process.
About the Speaker
Prof. Yin obtained his bachelor's degree in Civil Engineering from Zhejiang University (China) in 1997. He then worked as Engineer in Zhejiang Jiahua Architecture Design Co., Ltd. for five years. In 2003 and 2006, he obtained his master's and doctorate degrees in Geotechnical Engineering from the École Centrale de Nantes (France), respectively. Currently, Prof. Yin serves as the academic editor for “International Journal of Numerical and Analytical Methods in Geomechanics”, associate editor for 4 top-tier international journals: Geotechnique, European Journal of Environmental and Civil Engineering (EJECE), Geotechnique Letters, and ASCE-International Journal of Geomechanics. He is also serving as editorial board member of several journals, including the Canadian Geotechnical Journal (Can. Geot. J.), Acta Geotechnica, Transportation Geotechnics, Computers and Geotechnics, GeoRisk, etc..
Mostly recently, He funded a new journal “Marine Geotechnics”. Prof. Yin primarily engages in teaching and research in the fields of soil mechanics and geotechnical engineering. He has published over 300 SCI papers in leading international journals to date, with an H-index of 85 in google scholar. Prof. Yin's main research areas include: (1) Macro and micro characteristics of soil and its constitutive relationship, (2) Model testing in geotechnical engineering and large deformation numerical analysis, and (3) Application of artificial intelligence in geotechnical engineering.
